Bug in breakpoints: "!id_map.has(p_rid.get_data())" is true

[kzerot]

Godot version: 3.2.2.rc

OS/device including version: Linux Manjaro

Issue description:

It's floating bug, I have it only on more or less big project (on video - around 100 *.gd scripts)
When I'm trying to use breakpoints on certain places in a code debug console starting to spam error messages:

E 0:00:14.820   getornull: Condition "!id_map.has(p_rid.get_data())" is true. Returned: __null
  <C++ Source>  ./core/rid.h:160 @ getornull()

It can happen in different places of code. Also it happens not always.
https://youtu.be/QFScfQA6wNU

Steps to reproduce:
Really don't know. Make an big project
Minimal reproduction project:
Nothing

[SuperMeip]

This happens to me with some code without debugging as well :(

[Reneator]

Still getting this Error in 3.3.2 windows 10 with my big project. But this happens only (like reporter said) in certain places in the code, but when it does, it repeatedly spams those Errors (like a hundred per second) and slows down the editor drastically:

[TokisanGames]

RIDs that have already been freed are being sent to free again. You might be doing it in your code. The editor does it by itself due to a bug introduced as of Nov 2020.
You can read about it in #53374 and the fix is coming in PR #54650.

[akien-mga]

This may have been fixed by #54650 (included in 3.4.1 and 3.5). Please comment if you can still reproduce it.

[Torguen]

Still in 3.6 beta 1

[bspear8]

Observed in 3.5.1 stable

[hero-clashes]

It still exists in 3.5.3 stable

[thepunkoff]

This was happening in my 3.6.stable.mono project when I accidentally called some Godot API from a background thread, e.g. tried to instantiate some object on a scene from a network thread. This caused some unexpected behaviour and strange visual artifacts. As soon as I used calldeferred, the problem disappeared.